Definition
An experiment that removes or changes one component while holding others as constant as practical to estimate that component's contribution.
Distinguish it from nearby terms
Ablation estimates component contribution; a benchmark compares complete systems under a protocol.
Check your understanding
Interactions can make one-at-a-time ablations misleading, so report dependencies and repeated trials.